AI-Based SDLC Assistant with Multi-Agent Architecture

View at publisher Summarise with VIRAT
Authors: Niranjan P., S. Narain, R. Aishwarya
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Building software today feels like a continuous struggle to keep everything in balance; systems are getting far more complex, yet we're expected to deliver updates fast. While modern tools may help us to deal with the various aspects of project management and code generation, the early phases of the Software Development Life Cycle (SDLC), namely Requirements Gathering, Planning, and System Design. Continue to rely highly on manual effort and personal understanding. Limitations under this context can often result in confusing requirements, inconsistent documentation, and considerable rework in following development stages. To solve these problems, this research considers using Multi-Agent System (MAS), which has been designed with the idea of addressing some of the challenges by enabling Large Language Models (LLMs) to automate the critical early stages of SDLC activities. The proposed architecture uses domain-specific AI agents organized through the LangChain and LangGraph frameworks to autonomously perform requirement analysis, define the scope, and initial architectural modeling. Experimental results show that the multi-agentic system removes manual efforts by about fifty percent, while also achieving high clarity and consistency in the generated outputs. Results show us that the use of intelligent agent-driven workflows and procedures in the SDLC process can accelerate development timelines. They also improve the accuracy, ensure everything is complete, and maintain the reliability of the software engineering process.

Fields of Study & Subjects

Software Engineering Techniques and Practices Software Engineering Research Advanced Software Engineering Methodologies

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1624–1630
Publication date
21 January 2026
Language
EN